Video Multi-choice Question Answering on MSR-VTT 3k (val)
96.2AccuracyC4CL-mP
Evaluation Results
| Method | Links | |
|---|---|---|
| C4CL-mPEvaluation Protocol=FINE-TUNED, # params.=151M2023.04 | 96.2 | |
| VFCEvaluation Protocol=FINE-TUNED, # params.=164M2023.04 | 96.2 | |
| VFCEvaluation Protocol=ZERO-SHOT, # params.=164M2023.04 | 95.1 | |
| MMT (CLIP features)Evaluation Protocol=FINE-TUNED2023.04 | 95 | |
| CLIP-straightEvaluation Protocol=FINE-TUNED, # params.=151M2023.04 | 94.1 | |
| InternVideoEvaluation Protocol=ZERO-SHOT, # params.=≈ 460M2023.04 | 93.4 | |
| MMTEvaluation Protocol=FINE-TUNED2023.04 | 92.4 | |
| VideoCLIPEvaluation Protocol=FINE-TUNED2023.04 | 92.1 | |
| CLIPEvaluation Protocol=ZERO-SHOT, # params.=151M2023.04 | 91.1 | |
| ClipBERTEvaluation Protocol=FINE-TUNED2023.04 | 88.2 | |
| VideoCLIPEvaluation Protocol=ZERO-SHOT2023.04 | 73.9 |